vertex partition

[wang fangfang <ic...@gmail.com>]

Hi,If there is a graph in HDFS or HBase,then how it can be read into the
HAMA architecture,Is the BSPGroomServer read the data parallel or the
master read the data from the HDFS  and then send different part to
different BSPPeer(or BSPGroomServer).
Another question is How the vertex to be found in which BSPGroomServer 。Is
there a *synopsis data structure in the master.
Thank you very much！
*

Re: vertex partition

["Edward J. Yoon" <ed...@apache.org>]

Each GroomServer loads local data from HDFS, and then partitions the
graphs based on a partitioner. The default partitioner is
HashPartitioner, which finds a partition based on hash of the vertex
ID.

> Another question is How the vertex to be found in which BSPGroomServer

Based on hash of the vertex ID.

In HBase case, Range partitioning can be used instead of Hash partitioning.

On Sat, Dec 22, 2012 at 2:47 PM, wang fangfang <ic...@gmail.com> wrote:
> Hi,If there is a graph in HDFS or HBase,then how it can be read into the
> HAMA architecture,Is the BSPGroomServer read the data parallel or the
> master read the data from the HDFS  and then send different part to
> different BSPPeer(or BSPGroomServer).
> Another question is How the vertex to be found in which BSPGroomServer 。Is
> there a *synopsis data structure in the master.
> Thank you very much！
> *



-- 
Best Regards, Edward J. Yoon
@eddieyoon